LSTM-ED for Anomaly Detection in Time Series Data¶

In [ ]:
import time
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tensorflow as tf

from dataset import *
from plots import *
from metrics import *
from models_funtions import *

# Set style for matplotlib
plt.style.use("Solarize_Light2")

import plotly.io as pio
pio.renderers.default = "notebook_connected"
In [ ]:
# Path to the root directory of the dataset
ROOTDIR_DATASET_NORMAL =  '../dataset/normal'
ROOTDIR_DATASET_ANOMALY = '../dataset/collisions'

# TF_ENABLE_ONEDNN_OPTS=0 means that the model will not use the oneDNN library for optimization

import os
os.environ['TF_ENABLE_ONEDNN_OPTS'] = '0'

Variours parameters¶

In [ ]:
#freq = '1.0'
#freq = '0.1'
#freq = '0.01'
freq = '0.005'

file_name_normal = "_20220811_rbtc_"
file_name_collisions = "_collision_20220811_rbtc_"

recording_normal = [0, 2, 3, 4]
recording_collisions = [1, 5]

freq_str = freq.replace(".", "_")
features_folder_normal = f"./features/normal{freq_str}/"
features_folder_collisions = f"./features/collisions{freq_str}/"

Data¶

In [ ]:
df_features_normal, df_normal_raw, _ = get_dataframes(ROOTDIR_DATASET_NORMAL, file_name_normal, recording_normal, freq, f"{features_folder_normal}")
df_features_collisions, df_collisions_raw, df_collisions_raw_action = get_dataframes(ROOTDIR_DATASET_ANOMALY, file_name_collisions, recording_collisions, freq, f"{features_folder_collisions}1_5/")
df_features_collisions_1, df_collisions_raw_1, df_collisions_raw_action_1 = get_dataframes(ROOTDIR_DATASET_ANOMALY, file_name_collisions, [1], freq, f"{features_folder_collisions}1/")
df_features_collisions_5, df_collisions_raw_5, df_collisions_raw_action_5 = get_dataframes(ROOTDIR_DATASET_ANOMALY, file_name_collisions, [5], freq, f"{features_folder_collisions}5/")
Loading data.
Found 31 different actions.
Loading data done.

Computing features.

Progress: 0% Complete

0

Skipped feature extraction for pickFromPallet(1,2)=[true,1,0] 2022-08-11 14:37:37.436000 : 2022-08-11 14:37:37.421000.
Skipped feature extraction for placeToPallet(1,1)=[true,0] 2022-08-11 14:37:37.421000 : 2022-08-11 14:37:37.442000.
Skipped feature extraction for pickFromPallet(3,2)=[true,1,0] 2022-08-11 15:36:32.568000 : 2022-08-11 15:36:32.533000.
Skipped feature extraction for pickFromPallet(3,2)=[true,1,0] 2022-08-11 15:36:32.572000 : 2022-08-11 15:36:32.561000.
Skipped feature extraction for placeToPallet(1,3)=[true,0] 2022-08-11 15:36:32.533000 : 2022-08-11 15:36:32.572000.
Skipped feature extraction for placeToPallet(1,3)=[true,0] 2022-08-11 15:36:32.561000 : 2022-08-11 15:36:32.561000.
Features saved to ./features/normal0_005/features_statistical_200.0.csv.
--- 106.68732285499573 seconds ---
Loading data.
Found 31 different actions.
Loading data done.

Computing features.

Progress: 0% Complete

0

Skipped feature extraction for moveOverPallet(1,3)=[true,0] 2022-08-11 16:55:15.149000 : 2022-08-11 16:55:15.146000.
Skipped feature extraction for moveOverPallet(3,1)=[true,0] 2022-08-11 16:55:15.146000 : 2022-08-11 16:55:15.150000.
Features saved to ./features/collisions0_005/1_5/features_statistical_200.0.csv.
--- 38.67432403564453 seconds ---
Loading data.
Found 31 different actions.
Loading data done.

Computing features.

Progress: 0% Complete

0

Features saved to ./features/collisions0_005/1/features_statistical_200.0.csv.
--- 22.279465198516846 seconds ---
Loading data.
Found 31 different actions.
Loading data done.

Computing features.

Progress: 0% Complete

0

Skipped feature extraction for moveOverPallet(1,3)=[true,0] 2022-08-11 16:55:15.149000 : 2022-08-11 16:55:15.146000.
Skipped feature extraction for moveOverPallet(3,1)=[true,0] 2022-08-11 16:55:15.146000 : 2022-08-11 16:55:15.150000.
Features saved to ./features/collisions0_005/5/features_statistical_200.0.csv.
--- 19.869044065475464 seconds ---
In [ ]:
X_train, y_train, X_test, y_test, df_test = get_train_test_data(df_features_normal, df_features_collisions, full_normal=True)
X_train_1, y_train_1, X_test_1, y_test_1, df_test_1 = get_train_test_data(df_features_normal, df_features_collisions_1, full_normal=True)
X_train_5, y_train_5, X_test_5, y_test_5, df_test_5 = get_train_test_data(df_features_normal, df_features_collisions_5, full_normal=True)

Collisions¶

In [ ]:
collisions_rec1, collisions_init1 = get_collisions('1', ROOTDIR_DATASET_ANOMALY)
collisions_rec5, collisions_init5 = get_collisions('5', ROOTDIR_DATASET_ANOMALY)

# Merge the collisions of the two recordings in one dataframe
collisions_rec = pd.concat([collisions_rec1, collisions_rec5])
collisions_init = pd.concat([collisions_init1, collisions_init5])
In [ ]:
collisions_zones, y_collisions = get_collisions_zones_and_labels(collisions_rec, collisions_init, df_features_collisions)
collisions_zones_1, y_collisions_1 = get_collisions_zones_and_labels(collisions_rec1, collisions_init1, df_features_collisions_1)
collisions_zones_5, y_collisions_5 = get_collisions_zones_and_labels(collisions_rec5, collisions_init5, df_features_collisions_5)

RNN-EBM for Anomaly Detection in Time Series Data¶

In [ ]:
from algorithms.rnn_ebm import RecurrentEBM

# Disable eager execution
tf.compat.v1.disable_eager_execution()

classifier = RecurrentEBM(
    num_epochs=100,
    n_hidden=64,
    n_hidden_recurrent=32,
    min_lr=1e-4,
    min_energy=None,  # We'll set this to None initially and determine it after training
    batch_size=128,
    seed=42,
    gpu=None  # Set to None for CPU, or specify GPU index if available
)
# Train the RNN on normal data
classifier.fit(X_train)
print("RNN-EBM training completed.")
100%|██████████| 100/100 [00:10<00:00,  9.12it/s]
RNN-EBM training completed.

Predictions¶

In [ ]:
df_test = get_statistics(X_test, y_collisions, classifier, df_test, freq, threshold_type="mad")
df_test_1 = get_statistics(X_test_1, y_collisions_1, classifier, df_test_1, freq, threshold_type="mad")
df_test_5 = get_statistics(X_test_5, y_collisions_5, classifier, df_test_5, freq, threshold_type="mad")
Anomaly prediction completed.
Number of anomalies detected: 1 with threshold 16942358208.0, std
Number of anomalies detected: 112 with threshold 142.0788688659668, mad
Number of anomalies detected: 16 with threshold 632.2341003417969, percentile
Number of anomalies detected: 16 with threshold 612.8087134361267, IQR
Number of anomalies detected: 306 with threshold 0.0, zero

choosen threshold type: mad, with value: 142.0789
F1 Score: 0.8756
Accuracy: 0.9118
Precision: 0.8482
Recall: 0.9048
              precision    recall  f1-score   support

           0       0.95      0.92      0.93       201
           1       0.85      0.90      0.88       105

    accuracy                           0.91       306
   macro avg       0.90      0.91      0.90       306
weighted avg       0.91      0.91      0.91       306

ROC AUC Score: 0.9304
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image
Anomalies detected: 112
Best threshold: 120.1369 | F1 Score: 0.9067 | Precision: 0.8500 | Recall: 0.9714
Anomalies detected with best threshold: 120

	-------------------------------------------------------------------------------------

Anomaly prediction completed.
Number of anomalies detected: 1 with threshold 23346397312.0, std
Number of anomalies detected: 43 with threshold 109.4520492553711, mad
Number of anomalies detected: 9 with threshold 527.282246398925, percentile
Number of anomalies detected: 23 with threshold 197.76518487930298, IQR
Number of anomalies detected: 164 with threshold 0.0, zero

choosen threshold type: mad, with value: 109.4520
F1 Score: 0.8462
Accuracy: 0.9268
Precision: 0.7674
Recall: 0.9429
              precision    recall  f1-score   support

           0       0.98      0.92      0.95       129
           1       0.77      0.94      0.85        35

    accuracy                           0.93       164
   macro avg       0.88      0.93      0.90       164
weighted avg       0.94      0.93      0.93       164

ROC AUC Score: 0.9632
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image
Anomalies detected: 43
Best threshold: 120.1369 | F1 Score: 0.8767 | Precision: 0.8421 | Recall: 0.9143
Anomalies detected with best threshold: 38

	-------------------------------------------------------------------------------------

Anomaly prediction completed.
Number of anomalies detected: 10 with threshold 621.6637268066406, std
Number of anomalies detected: 10 with threshold 582.8217468261719, mad
Number of anomalies detected: 8 with threshold 679.8375244140625, percentile
Number of anomalies detected: 1 with threshold 742.22900390625, IQR
Number of anomalies detected: 141 with threshold 0.0, zero

choosen threshold type: mad, with value: 582.8217
F1 Score: 0.0303
Accuracy: 0.5461
Precision: 0.1000
Recall: 0.0179
              precision    recall  f1-score   support

           0       0.58      0.89      0.70        85
           1       0.10      0.02      0.03        56

    accuracy                           0.55       141
   macro avg       0.34      0.46      0.37       141
weighted avg       0.39      0.55      0.44       141

ROC AUC Score: 0.8445
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image
Anomalies detected: 10
Best threshold: 219.1636 | F1 Score: 0.8333 | Precision: 0.7237 | Recall: 0.9821
Anomalies detected with best threshold: 76

	-------------------------------------------------------------------------------------

In [ ]:
plot_anomalies_true_and_predicted(df_collisions_raw, df_collisions_raw_action, collisions_zones, df_test, title="Collisions zones vs predicted zones for both recordings")
In [ ]:
plot_anomalies_true_and_predicted(df_collisions_raw_1, df_collisions_raw_action_1, collisions_zones_1, df_test_1, title="Collisions zones vs predicted zones for recording 1")
In [ ]:
plot_anomalies_true_and_predicted(df_collisions_raw_5, df_collisions_raw_action_5, collisions_zones_5, df_test_5, title="Collisions zones vs predicted zones for recording 5")